#!/bin/sh
# ~jhs/public_html/bin/.sh/distfiles_cmpd	by jhs@
# Removes duplicate distfiles.
# I keep my distfiles in seperate smaller directories,
# segregated by release, with newest release directory containing most files,
# but a few older directories holding more stuff I need for legacy support.
# This shell script strips identical copies.

# See also: /site/domain/this/etc/make.conf.common
# See also: ~jhs/bin/.sh/distfiles_fetch

# Alternates I havent tried:
#	cd /usr/ports/sysutils/portupgrade ; make install ; rehash
#	  portsclean -D	 # Remove all distfiles not ref''d by ports
#	  portsclean -DD # Remove all distfiles not used by an installed port
#	cd /usr/ports ; make distclean
#		# Uses /usr/ports/Mk/bsd.port.mk	--distclean

bleat=true
bleat=echo	# Uncomment for Debug session.
DBG="distfiles_cmpd_debug: "
DBG=""

outer() {
	# $bleat "${DBG}Outer_0 $*"
	del=$1
	shift
	# $bleat "${DBG}Outer_1 $del : $*"
	for i in $*
		do
		# $bleat "${DBG}Calling Inner $del : $*"
		inner $del "$*"
		del=$1
		shift
		done
	}
inner() {
	# $bleat "${DBG}Inner $1 : $2 :"
	for i in $2
		do
		# $bleat "${DBG}zap $1 $i"
		zap $1 $i
		done
	}
zap()	{
	$bleat "${DBG}Stripping $1 against $2"
	( cd $1 && nice find . -type f -exec cmpd -v -d -s {} $2 \; )
	}

# Start Of Main.
D=/pub/FreeBSD/dists
dirs=
sense=`cd /usr/ports/distfiles;/bin/pwd` # not `pwd` as it returns distfiles
sense2=`basename $sense`
if [ "$sense2" = "distfiles" ]; then
	# You have a local distfiles/ to be stripped.
	dirs="${dirs} /usr/ports/distfiles"
	#	If this happens to be the same as another directory in the
	#	$dirs path, for each duplicate file cmpd will warn:
	#		...... share same device (major & minor) & inode ,
	#		link count is 1; Skipping.
else
	# You do not have a local distfiles/ to be stripped.
	# (Your /usr/ports/distfiles is presumably just a symbolic link
	# to eg $D/5.3-RELEASE)
fi

# Next directories are in strip order, ie first gets stripped against
# reference 2nd etc. Best in approx newest release last.
# In approx reverse order to /site/domain/this/etc/make.conf.common
# Not in simple 4 before 5 numeric order as releases 4.x 5.x & 6.x & 7.x
#	overlap / interleave in time.
# For release dates see http://www.freebsd.org/releases/index.html
#	/usr/www/en/releases

# dirs="${dirs} /chroot/usr/ports/distfiles"

cd $D
if test -d $D/current.old ; then
	dirs="${dirs} $D/current.old"
	# echo "$D/current.old exists so added to list ${dirs}"
else
	# echo "$D/current.old does not exist, so not added to list ${dirs}"
	true
fi
#	For when I occasionally manually:
#		cd /pub/FreeBSD/dists;mv current current.old;mkdir current
#		cd /pri/FreeBSD/branches/-current/ports
#		setenv PORTSDIR `pwd`
#		nice nice make -k -j 5 BATCH=yes fetch
#		distfiles_cmpd
#	so that current/ only contains what is now needed, & old files no
#	longer needed by current ports/ can be removed to save space.
#	Also where I can periodically
#		cp -R /usrb/chroot/usr/ports/distfiles

dirs="${dirs} $D/current"
#	current is moved about manually by an editor: sometimes
#	it is near the beginning, to keep it stripped
#	as small as possible, with most files in the latest release
#	directories. Other times, eg shortly before a new release is due,
#	to ensure I pre fetch as many current distfiles as possible I run
#	make fetch on current ports/, before that I either:
#		- move current to end of list to ensure I don't transiently
#		  have too many duplicate distfiles overflowing the disc.
#		- Or mv current current.old; make fetch ; distfiles_cmpd
#	Warning theres an additional pain: ports Mk/ macros are too dumb,
#	try to force me to remote refetch & reclick certain
#	license crap again, even if I already hold distfile localy in another
#	dir which is not directly /usr/ports/distfiles.

# /host/gate/usr/ports/distfiles is currently a link to $D/6.4-RELEASE
# so avoid doing it twice.
if false ; then
if test -e /host/gate/usr/ports/distfiles ; then
  dirs="${dirs} /host/gate/usr/ports/distfiles"
else
  true	# Dummy.
	#	Probably AMD has failed to mount, showable with:
	#	cd /host/gate
	#	/host/gate: Operation timed out.
fi
else
fi

# dirs="${dirs} $D/common"		# individual hosts'' local copies
	# Various hosts running different releases share this via sym links.

  dirs="${dirs} $D/4.11-RELEASE"	# 2005.01, 5.3 G

#	Shortly after a  release is made, I list the release last
#	& run a make fetch, as best time to grab stuff that may later migrate
#	elsewhere on the net & be harder to find.

  dirs="${dirs} $D/6.4-RELEASE"		# 2008.12 src tagged, 11.4 G

  dirs="${dirs} $D/7.4-RELEASE"		# tagged 2011.02.21, 400 K

  dirs="${dirs} $D/8.2-RELEASE"		# tagged 2011.03, 30 G & 1.2 G stripped

  dirs="${dirs} $D/9.0-RELEASE"		# tagged 2012.01

#  dirs="${dirs} $D/current"		# 86 G

  dirs="${dirs} $D/jhs"
	# Estic for which I am master repository.
	# Word perfect, that I'm licensed to use, & dont want stupid
	#	ports scripts forcing me to fetch again.

outer ${dirs}

# echo "$D/common/ should be empty, looking now:"
# ls $D/common

echo "Zero size files (if any) in /pub/freebsd/dists"
cd /pub/freebsd/dists
# find has no --exclude parameter so just list where I want to search.
#	Transient zero size files in ./httree/hts-cache/ref/*.ref
find [0-9]* current jhs -type f -size 0c | xargs rm